OpenAI gpt-5 Can Cook (If You Know The Trick)
OpenAI just dropped their GPT-5 prompting guide, and buried inside is "sauce worth millions of dollars." After reading their new cookbook, I discovered why GPT-5 feels lazy compared to Claude - we've all been prompting it wrong. In this live exploration, I dive into OpenAI's just-released documentation and uncover the hidden reasoning settings that transform GPT-5 from lazy to genius. Then I use it to expose a real protein powder scam that's been hiding in plain sight on nutrition labels. No fluff, just raw discoveries as we build a $100k app idea in real-time. What you'll discover in this video: ► The Cookbook: OpenAI's new prompting guide that just dropped (and why it's worth millions). ► Reasoning Effort: The hidden setting that makes GPT-5 think for 2 minutes straight. ► Agentic Persistence: How to make GPT-5 keep working until it completely solves your problem. ► The Protein Scam: Live demo exposing how nutrition labels lie about protein content using PDCAAS. ► Prompt Evolution: Watch me use Claude Opus 4.1 and Sonnet to refine prompts in real-time. ► Prompt IDs: OpenAI's new feature that protects your system prompts from being stolen. ► Live Build: Creating a "Truthful Protein" app that could genuinely help millions. Key revelations from OpenAI's guide: - Why setting reasoning to "low" instead of "high" often works better in Cursor - The exact prompt to make GPT-5 persistent: "keep going until completely resolved" - How GPT-5's tool calling differs from Claude (and why it matters) - The mathematical calculation that reveals your protein powder's real value 🕒 Key moments: 00:00 GPT-5 prompting is the new fire 00:24 Why GPT-5 can feel "lazy" 00:42 New guide: Controlling the model's effort 00:53 Exploring the OpenAI Cookbook 02:20 Controlling "agentic eagerness" 03:59 Real-world example: The protein scam 05:44 Prompting GPT-5 to "think hard" 06:55 "Aha!" moment: Revealing the real protein value 07:29 The "Truthful Protein" app idea 08:24 Using Claude to refine the new prompt 09:16 Sonnet vs. Opus: Who writes better prompts? 10:12 Building the final, comprehensive prompt 12:10 New feature: Using a prompt ID 13:30 Live test: Analyzing a nutrition label image 14:17 The power of learning by doing Join 160+ builders in our private Discord where we ship AI apps together: https://rfer.me/community 🎯 Get weekly insider AI updates: https://dub.sh/RayMasterAI 👉 Follow Me: Ray on X: https://x.com/RayFernando1337 My AI Resources & Blog: https://rayfernando.ai #gpt5 #chatgpt #openai #aicoding #promptengineering #cursor #claude #aiagents #buildsinpublic #cookbook #gpt5cookbook
